On the initialization of tropical cyclones with a three dimensional variational analysis
A method of initializing tropical cyclones in high-resolution numerical models is developed by modifying a data assimilation system, the NRL atmospheric variational data assimilation system (NAVDAS), which was designed for general mesoscale weather prediction using a three-dimensional variational (3DVAR) analysis with intermittent updates. The method includes the following three upgrades to overcome difficulties resulting from tropical cyclone initialization with the NAVDAS analysis. First, synthetic observation soundings are generated on 9 vertical levels at 49 points for strong storms (v max > 23.1 m s−1) and 41 points for weak storms around each cyclone center to supplement the observations used by the analysis. Secondly, a vortex relocation method for nested grids is developed to correct the cyclone position in the background fields of the analysis for each nested mesh. Lastly, the 3DVAR analysis is modified to gradually reduce the horizontal length scale and geostrophic coupling constraint near the center of a tropical cyclone for minimizing the problems introduced by improper covariances and coupling constraint used in the analysis. The synthetic observations significantly improve the intensity and structure of the analysis and the track forecast. The vortex relocation significantly improves the first guess background, avoiding the large analysis corrections that would be needed to correct cyclone position, and reducing the imbalance introduced by such large analysis increments. The modifications to the analysis length scale and geostrophic coupling constraint successfully improve the inner core analysis, providing a tighter circulation, and reducing the underestimate of the mass field gradient. Among the three upgrades, the vortex relocation provides the largest improvement to the tropical cyclone initialization and forecast.
KeywordsTropical cyclone initialization Synthetic observations Vortex relocation 3DVAR analysis for tropical cyclones
This research is supported by the Office of Naval Research through program PE-0602435 N. We benefit from the discussions with the rest of COAMPS-TC development team members, especially Drs. James Doyle and Richard Hodur. We also appreciate comments and suggestions from two anonymous reviewers.
- Daley R (1991) Atmospheric data analysis. Cambridge University Press, Cambridge, p 457Google Scholar
- Daley R, Barker E (2001) The NAVDAS source book 2001. NRL/PU/7530–01-441. Naval Research Laboratory, Monterey, p 163Google Scholar
- Liu Q, Marchok T, Pan H, Bender M, Lord S (2000) Improvements in hurricane initialization and forecasting at NCEP with global and regional (GFDL) models. NCEP/EMC Tech Proc Bull 472:7Google Scholar
- Liu Q, Surgi N, Lord S, Wu W-S, Parrish D, Gopalakrishnan S, Waldrop J. Gamache J (2006) Hurricane initialization in HWRF model. 27th Conference on hurricanes and tropical meteorology, Monterey, CA, Amer. Meteor. Soc., P5.13Google Scholar
- Pauley PM (2003) Superobbing satellite winds for NAVDAS. NRL/MR/7530–03-8670. Naval Research Laboratory, Monterey, p 102Google Scholar